fix: update sample rate

[AlexanderSutul]

Fix for this issue:
400 sample_rate_hertz (48000) in RecognitionConfig must either be omitted or match the value in the WAV header (44100).

[C-Pro]

Can you also force output sample rate (resample) in ffmpeg for both video and audio?

[C-Pro]

Most likely the flag is -ar 44100

[egregors]

Looks like the conceptual problem here, a bit bigger than it seems like. When I proposed to think about degradation, I mean the followings:

1. Let's say we have an abstract functionality: nastya-mode
2. This functionality is a composition of smallest functions: find in the chat right messages, remove messages, ban bastard, translate audio to text, post recognized text.
3. And we here the pretty explicit priority: catch msg -> delete -> ban -> audio-to-text
4. A degradation of some functionality composition is saving main functions even if other parts are not working properly. In other words, when some part of the composition isn't available, we are degrading the whole functionality by excluding \ skipping broken parts. But the main func still must be done. This is the general idea.

I believe it will be good, if we make a little refactoring to support this mechanic.

So, in our particular example, nastya-mode must in any cases perform catch -> remove -> ban parts, and if it possibly – trying to convert audio to text.